Presenting Data: Graphs and Graphics Idea Book 3
This Idea Book stems from a series of science communication workshops conducted in Thailand in September 2017, and is intended as a tool for researchers who participated in these workshops and for other scientists wishing to improve their communication skills. The Thailand workshops were co-organised by EURAXESS ASEAN and Thailand’s National Science and Technology Development Agency (NSTDA) with the support of the ThailandEuropean Union Policy Dialogues Support Facility
Author: Sanna Fowler Content from this idea book may be used for non-commercial, educational purposes only. 2
Contents
Introduction
5
Deciding what to present
6
What kind of graph
14
Understanding figures
26
Infographics
28
Some useful tools
30
Summary
31
3
4
Introduction As scientists, your world revolves around data - it’s pretty much the basis for everything you do. When you write a paper, your data (the results you obtained through experimentation) are the central part, with your explanations and conclusions based on these observations. When communicating to a non-expert audience, it’s completely the opposite! You can use data as a tool to illustrate your message (among other tools like images, film etc) but it’s not the focal point. The aim is not to explain your results. This is a really hard thing to do – you’ve slaved hard for this data, worked weekends, run replicate experiments - maybe you have the most beautiful graphs. It’s going to be painful, but you have to face up to the fact that you’re not going to be able to show it all! The good news is that a good graph or graphic can be one of the most powerful things you can put in your presentation. It’s visual, it enhances your credibility and is easier for people to remember (if done well). If done badly, it can be the quickest way to turn off an audience…so how to tackle graphs? 5
1. Sort through your data
It’s likely that you have masses of graphs you could show in a presentation but the first step is to sort through and decide which ones are going to help you make your point(s). Remember, you’re not making a data-driven presentation like you would for a scientific audience but rather using data, as and when it's needed, to illustrate your point. You will probably need to pool data from different sources and turn your scientific graphs into clear images that people will easily grasp.
6
Image: visual.ly
2. Parse & Filter
Remove any data that ‘clutters’ your argument or that takes people away from the point you’re making. For a non-expert audience you don’t need to be as rigorous with showing your controls as you would for an expert audience. Keep the data you want to compare eg. A vs B or A vs control but eliminate other information
1 message = 1 graph If you want to make another point – separate it out
Image: visual.ly
7
3. Focus on your message
The message you’re trying to make should leap out at you, without the need for too much explanation. You can use text, colour and even pictograms to augment your graph and make it clearer.
8
Image: visual.ly
We’ve all sat at the back of an auditorium at some point and seen presentations where the researcher has copied their graphs from a paper or excel. Typically you can’t see a thing, and this is usually made worse by cramming 3 or 4 graphs on the same slide!
4. Scale everything up
When you get your data ready for presentation, think about that person right at the back and follow these basic rules:
● Increase the font size of your axes, title, labels – min 20 point; ● Increase the weight of your lines; ● Add colour – max 3 including a neutral (black/grey etc) Use ppt/keynote to help you pick colours that go together – choose a theme; ● If possible don’t have a legend – include it in the graph itself; ● Add a text callout/explanation that highlights the point you’re trying to make. Lastly put your computer on a table take a good 5 paces back and see if you can understand the graph.
Image: visual.ly
9
5. Order your data If you have a number of data points for discrete values, make sure you order them sequentially from smallest to largest (or vice versa): Use colour to highlight the particular points you’re interested in. If you’re trying to highlight a trend (for example in the lower graph), order and annotate your data so this is clear. Energy production from wind turbines (Germany, 2012)
10
6. Create AND Curate Scientific presentations are all about your results and your data. But if you want to tell the bigger picture, to a wider audience, you need additional information. This may mean using other sources to illustrate your point. Just make sure you reference your source.
11
“Not everything that can be counted counts� William Bruce Cameron, 1963
12
This is a shorter version of the full quote: ”not everything that counts can be counted and not everything that can be counted counts.” As a scientist you’re used to counting everything, but remember when it comes to making a presentation, not all of your data counts! So assuming you’ve now decided what data you need to show to make your point, how should you show it?
13
1. Scatter plots and line graphs
These graphs are pretty ubiquitous in science presentations. If you’re going to use this kind of plot remember the following:
1. Make your points bigger 2. Use clear contrasting colours 3. Clearly label your data series on the graph (avoid using a separate legend that your audience have to refer to). 4. Ideally only use 2 data series (comparing A to B for example). Bear in mind that the message you’re trying to convey from your graph should be simple and evident (too many data series will confuse your message).
14
Another staple in scientific reporting, bar graphs can be vertical or horizontal. Horizontal graphs work well for x values like distance, time or length that we visualise more easily horizontally.
2. Bar Graphs
For a vertical graph, try to limit the number of bars to 3 or 4 – if you’re just trying to make 1 point, in theory, you only need 2 bars (A compared to B)
This was taken from the guardian datablog (a wellrespected UK media) but the following pages outline how we could improve this graph for a non-expert audience…
Source: The Guardian 15
Improving Bar Graphs Here’s the same data from the last graph but pared down to the essentials. If we imagine that the primary message we want to get across is that there are more men than women in STEM careers, we can reduce the information to a single year (since there is no great improvement between years – this could be a secondary message). The graph on the left is a simpler version of the original, with bigger labels and a ‘call out’ added to make the point that there are seven time more men than women. The graph on the right takes things further – it doesn’t even look like a graph! By using pictograms or shapes and scaling their volumes correctly, the message is immediately clear. For non-expert audiences, graphs can be daunting – even ones that are easy to understand. By using shapes or images, you can take out some of the ‘fear factor’ for your audience and put them more at ease. It’s your responsibility to help them understand.
16
Number of women and men in STEM occupations
7x 4.4M
More men than women in STEM careers in 2012
0.6M
17
17
Horizontal Bar Graphs Horizontal bar graphs can be useful if you have several data series or conditions, and they make a trend more easy to see. Here again, you can simplify things as much as possible to help your audience. A clear title and labels are enough – you can even do away with the axes.
18
5 most common cancers in women Breast Lung Bowel Uterus Malignant Melanoma
19
0
10K
20K
30K
40K
50K
60K
19
3. Pie Charts vs Donuts Pie charts have been used for centuries (the first by Scottish engineer William Playfair in 1801), and they are an extremely effective way to show percentages or parts of a whole.
label 12% 16%
72%
total 72%
20
Consider using donuts instead of pie charts where possible – the main advantage is that they give you space in the middle to add information. This could be your takehome message or a pictogram as on page 8.
There are a few rules to consider when using pie charts or donuts: 1. Place your largest values at the top of the image, and the smallest at the bottom. A more usual method is to order your data clockwise, with the biggest data set starting at 12 o’clock and the rest following on. 2.
Don’t compare two pie charts or donuts – if you’re comparing percentages for two data sets, use stacked bar graphs instead
1.
2.
3. Avoid using more than 4 or 5 data points. Group the smaller ones as ‘other’ if you can.
Image: visage.co
21
4. Tree plots MOOCs courses taken by African students
Tree plots can be used instead of pie charts/donuts when you have a large number of groups. The difficulty with this kind of plot is that it can be difficult to precisely compare similar values. But it is a very effective visual tool if you want to give an overall impression or if you have one stand out group that is bigger than the others.
22
The bubble plot is a great tool for presenting data – in this case replacing a bar graph. It’s extremely visual and immediately gets the message across, providing there are not too many groups. Again, it can be hard to precisely compare similar sized bubbles but this format leaves enough space to add values if needed. A bubble plot allows you to plot up to 5 different values for a given point in your data. For a great example see Hans Rosling’s presentation for the BBC. https://www.youtube.com/watch?v=jbkSRLYSojo
5. Bubble plots
5 most common cancers in women
Lung Bowel
Breast Melanoma
uterus
23
Understanding figures There has been a lot of research that shows that both expert and non-expert audiences understand statistics better if expressed as a frequency (eg. 1 in 5) rather than as a percentage (eg. 20%). See here for an example. So, if possible, use frequencies wherever possible and visually is even better.
VS
24
Avoid making your audience work hard for information‌most will simply turn off. If you can: 1. Relate numbers to real things people can picture rather than an abstract number For example: 4 x height of Everest rather than 35,392m ..fit 4 hundred on a pinhead rather than 0.3mm 3 times the speed of sound rather than 1000 m/s
2. Take out the technical where possible For example: 26 out of 1000 vs 2.6% 1 million vs 1 x 106 (avoid exponentials wherever possible) 25
Infographics Infographics are a form of data visualisation that have taken off over the last decade and in a world of big data, can be increasingly complex and beautiful. The essential point of infographics is that they are a visual form of storytelling – taking your audience through your messaging using a combination of: 1. Data 2. Images or pictograms 3. Text You don’t have to be a graphic designer or create complex images – there are a number of online tools outlined in the following slide that can be used (often for free) to help you make your data more visual.
26
Image: technologist.eu
27
Infographics Tools ●Infogr.am/ ●https://mindthegraph.com/ ●https://www.canva.com/ ●https://www.edrawsoft.com/infographics/index.php ●https://plot.ly/ ●http://piktochart.com/ ●http://www.easel.ly/ ●http://visage.co/ Inspiration ●http://www.theguardian.com/data ●http://www.nytimes.com/upshot/ ●http://visual.ly/ 28
In summary: Take the work out of data for your audience:
1 figure = 1 message The less it looks like a graph, the better! Use text callouts to give your key message Use visual storytelling/infographics
29
About the Workshop Organisers EURAXESS-Researchers in Motion is a unique pan-European initiative delivering information and support services to professional researchers. Backed by the European Union and its Member States, it supports researcher mobility and career development, while enhancing scientific collaboration between Europe and the world. For more information on EURAXESS please contact Simon Grimley and Susanne-Rentzow-Vasu at asean@euraxess.net
The National Science and Technology Development Agency (NSTDA) is a semi-autonomous government agency under Thailand’s Ministry of Science and Technology. NSTDA is committed to achieving four strategic missions comprising: 1) research & development 2) technology transfer 3) human resource development and 4) S&T infrastructure development, through its four main national research centers; the National Center for Genetic Engineering and Biotechnology (BIOTEC), the National Metal and Materials Technology Center (MTEC), the National Electronics and Computer Technology Center (NECTEC), and the National Nanotechnology Center (NANOTEC). For more information on NSTDA please contact Phawika Rueannoi at Phawika.ruennoi@nstda.orth
The Thailand-European Union Policy Dialogues Support Facility (PDSF) is a programme funded by the European Union to further advance Thailand’s cooperation with the EU towards achievement of sustainable economic and social development goals and global integration.
30
Author: Sanna Fowler
30